How to Check the Date and Version of a Fund Document

Every fund document carries a date on its face, and the date decides what the document can still be trusted for. Find the stated date, establish which kind of date it is, then sort the fields into three bins: the ones fixed until the document is revised, the ones revised on a cycle, and the ones true only at one stated moment.

Those four moves are the whole procedure, and the six steps that produce them follow. Here is what it rests on. Several documents exist about one scheme. The documents are different lengths, they were published at different moments, and they get changed at different moments after that. So the useful question is never whether a document is good. The question is whether the copy in hand is current for the particular field about to be read.

One scheme runs through every step. The documents worked through below belong to the Girnar Large Cap Equity Fund, an open ended equity scheme run by Girnar Asset Management Limited, an invented firm, carrying net assets of Rs 4,200 crore, 120.00 crore units outstanding, 3,80,000 folios and an expense ratio of 1.65 per cent of the scheme's assets a year. Its portfolio is managed by Kalyani Bhagat, and operations there are headed by Sohail Merchant.

Think about a printed railway timetable for a moment, the kind pinned up at a small station. The line it serves does not change. The names of the halts do not change. The departure times change quietly whenever somebody reissues the sheet, and the platform number changes on the day. Three different things on that one sheet age at three different speeds, so a reader asking whether the timetable is out of date has asked an unanswerable question. A fund document behaves the same way, and the procedure below is built around that single fact.

Where does a fund document state its own date?

On its face, and there are only three places worth searching before it can be called absent. The cover or the opening sheet, where a document usually states what it is and when it was put out. The running header or footer that repeats on every sheet, the usual place for a publisher's edition line. The very end, where a document is signed off or stamped. If the date is in none of those three places, it is absent, and hunting for it in the body of the text is where an hour disappears.

The harder half of this step is knowing what does not count. The date a browser recorded when it saved the file does not count. The date a search result displayed beside the link does not count. The date a computer shows as last modified does not count, and that one is the most convincing of the three. A last-modified date looks like a property of the document when it is a property of the file sitting on the machine. All three describe the copy rather than the document, and a copy has a history of its own that has nothing to do with when the publisher put the document out.

Consider a photocopy of a rent agreement. The shop made the copy on a Tuesday and the machine printed the day and time along the edge. Nothing about that stamp says when the agreement was signed, when it was last amended, or whether the landlord has since issued a fresh one. Nobody confuses the two on paper. On a screen the two collapse into one line of small grey text, and that is the entire reason this step exists.

Sometimes there is genuinely nothing. A document can carry no date of its own anywhere on it. An absent date is not a failed step but a result, and the absence gets written down as a finding and carried forward to the end of the procedure. A recorded blank is worth something to the next person. An unnoticed blank is worth nothing to anybody.

Which kind of date is it, and why does the kind matter?

Because five different dates can appear on a document and each one answers a different question. An issue dateThe date a document was first published, before any later change to it. says when the document was first put out. A revision dateThe date the text of a document was last changed by its publisher. says when its text was last changed. An as-at dateThe moment at which a figure printed inside a document was true. says at what moment a figure inside it was true. A period-end dateThe close of the window that a reported figure covers, such as the last day of a reported year. says which window a reported figure covers and where that window closed. And a printed-on date says when this particular copy came off a machine, a fact about the copy rather than about the document.

A document can carry several of these at once, and the step is not finished until it is clear which date governs which part of the document. That sounds fussy until the case it guards against turns up. A document revised in one month can carry a size figure that was true in an earlier month and a return figure covering a year that closed earlier still. Three dates, one document, and only one of them has anything to say about any given number in it. Reading the wrong one makes a figure look fresh because the text around it was recently retyped.

How is the version in hand identified?

The search is for a version markerWhatever a publisher uses to tell one edition of a document apart from another, such as a number or a revision line., and a marker takes one of four shapes. The marker can be a plain number or letter, printed as an edition line. A revision date often does double duty as the marker, and that is why naming the kind of date comes first. The marker can be a reference to an addendumA notice issued by a publisher that changes something in a document already published., a separate notice that changes something in a document already published, in which case the document in hand is only current when read together with every notice attached to it. Or there can be no marker at all. An absent marker is the fourth shape and the one people handle worst.

When there is no marker, the step does not end in a shrug. The step ends at the publisher's own current copy. Girnar Asset Management publishes the documents for the Girnar Large Cap Equity Fund on its own site, and the copy sitting there is the copy the publisher is standing behind today. A second route runs at the industry level, through the Association of Mutual Funds in India (AMFI) at amfiindia.com. The marks of a current version, and the places a current version has to be obtainable from, are set by rule and are read at sebi.gov.in. A requirement of that kind changes, and a reproduced copy of one does not merely go out of date, it goes wrong.

Which parts of a document can go stale, and which cannot?

Sorting the contents is the step that decides the whole check, and it is the step people skip. A document does not go out of date as a single object. Its fields do, at three different speeds, so the output of this step is a marked-up document rather than a verdict on it. Run a bin sortThe habit of grading each field in a document by how quickly that field goes out of date, rather than judging the document as a whole. over the contents and put every field into one of three bins.

Bin one holds the fields that stay fixed until the document itself is revised: what the scheme states it is trying to do, how it states it will invest, and the plan structure it offers. Bin two holds the fields that are revised on a cycle or changed by a notice: the charges, the personnel, the terms. Bin three holds the fields that were true at one stated moment and at no other: the size, the value per unit, the holdings, and a return for a window that has closed. A field in bin three was already out of date on the day the document was printed, and that is not a defect in the document but the nature of the field.

Here is the household version. Look at a school notice board. The school's name and address sit at the top and change effectively never. The list of staff and the fee schedule change when somebody puts up a fresh sheet. The lunch menu and the day's absentee list are true for one day and nobody pretends otherwise. Same board, same drawing pins, three speeds. Nobody standing in front of a school notice board asks whether the board is current, and nobody should ask it of a fund document either.

What do the three bins hold once the sorting is actually done?

Take the bin two field first. A reader is most likely to act on it. The Girnar Large Cap Equity Fund's expense ratio is 1.65 per cent of the scheme's assets a year. On net assets of Rs 4,200 crore that is Rs 69.30 crore a year for the scheme as a whole. Across 365 days it comes to about Rs 0.1899 crore a day, and across 120.00 crore units to about Rs 0.001582 per unit a day. Bring it down to the scale of one holder and it becomes concrete: Rs 4,200 crore spread over 3,80,000 folios is an average folio of about Rs 1,10,526/-, and 1.65 per cent of that is about Rs 1,824/- a year.

Every one of those figures is worth exactly as much as the ratio behind it, and the ratio sits in bin two. The arithmetic is not the fragile part of that calculation; the field it started from is. The chain is worth working in both directions before it is trusted.

The last two rows say something the first six cannot. The two backwards routes do not land on the identical figure, and the reason is rounding rather than error. The average folio is exactly Rs 1,10,526.32 and was rounded down to Rs 1,10,526/-. The charge on it is exactly Rs 1,823.68 and was rounded up to Rs 1,824/-. Multiplying the rounded-up figure back across 3,80,000 folios gives Rs 69.312 crore instead of Rs 69.30 crore, a difference of about Rs 0.012 crore that was created entirely by that one rounding and by nothing else. Two of the other figures were rounded too: about Rs 0.1899 crore a day is Rs 0.18986 crore rounded up, and about Rs 0.001582 a unit a day is Rs 0.0015822 rounded down.

Why does a closed period figure never go stale and never update?

Because the window it describes has shut, and nothing that happens afterwards can reach back into it. The Girnar Large Cap Equity Fund returned 13.4 per cent net for one stated year, measured value per unit to value per unit, and net there means the figure is already after the scheme's own charge rather than before it. The stated benchmark returned 12.1 per cent for the same year, and nobody is billed to hold a calculation, so that figure carries no cost at all. Both bases are named because a return without its basis is not a usable number.

A return for a window that has shut is a closed periodA window that has already ended, so the figure describing it can never change and never refreshes either. figure, and it behaves unlike anything else in the document. The figure will still be correct in ten years. The same figure will also say nothing whatever about the ten years in between. A reader holding it a year later holds something permanently right and almost entirely useless. A permanently right number is a stranger object than a wrong one, and much harder to notice. It sits as a special case of the fastest bin rather than as a fourth bin of its own: it belongs with the moment-only fields because it describes one closed window, even though it never refreshes the way the others do.

Two cautions travel with that pair of figures. The first is the difference between them. Subtracting 12.1 from 13.4 leaves 1.3 percentage points. The 1.3 is the arithmetic difference between two returns rather than a return of 1.3 per cent in its own right, and the two read almost identically in print. The second caution is bigger. 13.4 per cent is a figure after the scheme's charge and 12.1 per cent is a costless basisA basis carrying no charge at all, because an index is a calculation that nobody is billed to hold.. Subtracting one from the other is not a like for like comparison at all. Putting both sides onto one basis is a different job with its own arithmetic, and it is covered separately rather than attempted here.

How is a copy checked against the publisher and the regulator?

Three destinations, in order, each of them for a different thing. The first is the publisher's own site. Girnar Asset Management keeps the current documents for the Girnar Large Cap Equity Fund there, along with any notice that changed one of them. The second is the Securities and Exchange Board of India (SEBI) at sebi.gov.in, the source of the requirements behind all of that. The third is AMFI at amfiindia.com, the industry level route to a scheme's published documents, and it is a route rather than a rule maker.

The check against the publisher has three destinations and the bin sort had three bins, and those two threes have nothing to do with each other. The counts collide by coincidence. A mental link between them leads to a hunt for one destination per bin, and no such destination exists.

Every one of those requirements is set by rule and every one of them moves, so the step names where to look and states nothing about what has to be there or for how long. That is the discipline the whole procedure runs on. A reproduced revision cycle is helpful right up until the cycle changes, and from that day forward it is confidently wrong to every reader who trusts it.

What gets written down when a check comes back unresolved?

Both halves get written down: what the check established, and what it could not. A version confirmed, a date confirmed, a bin sort completed, and then every unresolved rowA check written down as unfinished, with the reason, rather than left blank or quietly dropped. with its reason beside it. A check that leaves no written record cannot be repeated and cannot be handed to anybody else, including the person who ran it three months ago. The written output is the actual product of the procedure. Everything before it is just reading.

Two unresolved rows come out of this scheme's own record, and they are worth seeing side by side because they fail for different reasons. The first is whether a notice has changed the expense ratio of 1.65 per cent since the copy in hand was published. The copy carries no signal either way, so that row stays unresolved. The second is whether the stated benchmark's 12.1 per cent for the stated year is computed on price alone or with income reinvested. This record simply does not say, and a scheme return includes income the scheme received. Neither row gets a guess written into it.

One more collision worth naming before this step closes. The card has four empty rows, and the closed period figure was described earlier as a fourth kind of field. The two fours are unrelated. One counts rule-set rows on a form and the other counts kinds of content in a document, and reading a correspondence between them would invent a structure that is not there.

Can one copy be current and out of date at the same moment?

Yes, and that sentence is the whole procedure compressed into one line. One copy of one document, at one reading moment, can be perfectly current on what the scheme states it is trying to do, a notice behind on its charges, and well past the moment its size and value per unit were true, all at once. Nothing about the copy changed; only the field being read changed. Judging a document as a single object throws away the only distinction that matters.

Girnar Asset Management Limited, the Girnar Large Cap Equity Fund, Kalyani Bhagat and Sohail Merchant are invented.
Educational material. Not advice on any investment, tax, budget or market position.